Based on these consideration to select neutral wind profiles the following conditions have been imposed: 1. The presence of neutral or quasi-neutral conditions: z/|L| <0.01 2. The vertical stationary of wind direction profile (RWD = WD/WD mean ≤ WD lim ) 3. The constancy of vertical profiles of u* (R u* = u*/ u*mean ) where WD is the wind direction.
As known, the neutral profiles is characterized by the logarithmic equation that is strictly valid in flat homogeneous terrain. For all data set, we analysed different combination of above criteria and assuming |z/L| < 0.005, R(u*)=0.15 and RWD < 5% finally 288 profiles representative of the main neutral conditions measured at Ny-Alesund, were selected. .
2.2 Choice of output variables as derived from best fit process The logarithmic wind profile equation permits the calculation of z0 and k starting from measured values of U(z) and u* for the selected 288 profiles. In our wind profiles analysis, we have used the four cup anemometers for wind data and the sonic at 3.7 m for the evaluation of u*. To compute z0, 2 simulations were carried out: • classic case (Clas): is by using all the four of cup anemometers . • unconventional case (Fit): the values of z0 is determined from the fitting process using all anemometers . As performance index for each profile, the percentage of reproducibility Rp of each wind measurement averaged for all the heights has been evaluated.
2.3 Analysis of wind sectors The selected 288 profile are distributed within the three main sector that identify the wind field a Ny Alesund: 45-135°, 180°-270° and 270-360°. The number of profiles for the selected sector are 120 from the NE-SE, 23 from SW and 100 from NW directions .NE MANCANO 45 !!!! The wind from NW sector is the colder one (T=-5.33°C), and this is coherent with the orography of the area and the local distribution of the mountains and valley. For the data selected, the minimum speed is observed from SW, where the wind overpass the mountain before to reach the site of measurement.
3. Results The best wind speed profiles are reproduced applying the fitting process to all the sensors heights and calculating both z0 and k. The values of Rp for the three wind directions are given in table 1. Note that the best wind reproducibility is obtained using all profiles and leaving system free to calculate k and z0.
RP(Fit) RP(Clas) SE 0.73 1.72 SW 1.74 8.04 NW 0.91 2.55 Table 1: average Rp values for the three main wind directions
The values of z0 calculated with different formulations are consistent with the characteristics of the site, but they are very different and depend by the used method. In Table 2 the average values of z0 in mm for the three directions are given. An anomaly can be noted in the SW direction: when the Fit method is adopted the value of z0 is smaller than for other sectors; in the case of classic formulation ( Clas) the situation is reversed and value of z0 (70.5mm ) is out of the expected range of variability expected.
z0 (Fit) z0 (Clas) SE 1.27 3.02 SW 0.67 70.45 NW 1.04 3.02 Table 2: z0 ( in mm) from different methods and wind directions
The method based using all anemometers b) is also the most adequate to reproduce the observed wind values.
4. Conclusions These results provide a clear indication that the values of z0 dependent by the formulation chosen in the calculation. The analysis performed provides results whose significance should be thoroughly investigated.